Novel Approach for Vehicle-Self-Localization

This paper presents a novel approach for dead reckoning. The described localization system consists of an inertial navigation system (INS), a magnetic, angular rate, gravity sensor (MARG sensor) and an odometry an odometry model. In contrast to conventional odometry models, a kinematic two-track model is implemented. The odometry model uses wheel-individual steering angles. It can therefore be applied for vehicles with all-wheel steering and steering geometries that allow opposite steering angle directions at one axle. An error-state Kalman filter is used to merge the individual submodels. While conventional odometry based localization systems only consider the vehicle longitudinal and lateral speed and position change, the proposed localization system is also able to correctly represent movements along the vehicles vertical axis. Furthermore, the algorithm uses the vehicle acceleration calculated from the odometry model to increase the robustness of the orientation estimation. The aim of the localization method is a high positioning accuracy in the low speed range, e.g. during precise maneuvering or parking. For this reason, the accuracy of the localization system is demonstrated by driving tests on a parking lot. The all-electric vehicle platform flexCAR is used as a test vehicle. Through its symmetrical design, this vehicle is able to realize wheel steering angles of up to 30° on the front and rear axles. Due to its maneuverability it is particularly suitable for the investigation of parking and (maneuvering).